Intelligent Motion Control Method Based on Directional Drive for 3-DOF Robotic Arm

There are observations indicating that the primary motor cortex of the brain drives arm movement by encoding directional signals. In this study, we proposed an intelligent motion control method based on directional drive for target reaching task of robotic arms. Our proposed method focuses on simulating the control mechanism above by the direction-based reward function in reinforcement learning (RL). Specifically, this reward function was designed using the deviation between the actual motion direction and the expected direction of the robotic arm. Then, an intelligent motion control algorithm established by deep deterministic policy gradient (DDPG) was put forward to enable the end-effector of the robotic arm to reach the target automatically. Finally, our presented method was verified by simulations on 3-DOF robotic arm, and the success rate of target-reaching is around 90%. The experimental results show that our designed reward function effectively simulates the control mechanism of the brain.